OpenAI GPT OSS 120B: LLM Security & Jailbreak Resistance
As of 2026-07-04, OpenAI GPT OSS 120B has an attack success rate (ASR) of 15.0% on Guardion's LLM Vulnerability benchmark — ranking #15 of 93 models. It is strongly resistant to jailbreaks, with occasional bypasses under sustained attack.
Attack Success Rate (ASR) is the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks (HarmBench framework). Lower is safer. "Est." models are calibrated from public safety evaluations pending a Guardion benchmark run.
How GPT OSS 120B compares
Frequently asked questions
How secure is OpenAI GPT OSS 120B against jailbreaks?
OpenAI GPT OSS 120B is strongly resistant to jailbreaks, with occasional bypasses under sustained attack. On Guardion's LLM Vulnerability benchmark it records a 15.0% attack success rate (85% robustness), ranking #15 of 93 models (estimated from public safety evaluations).
What is GPT OSS 120B's attack success rate (ASR)?
GPT OSS 120B's overall ASR is 15.0% — the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks. Lower is safer.
Is GPT OSS 120B more secure than GPT-5.2 Pro?
GPT-5.2 Pro is more robust: 3.8% ASR vs GPT OSS 120B's 15.0%. See the full head-to-head comparison for the per-attack breakdown.
How can I make GPT OSS 120B safer to deploy?
Even robust models are bypassed under sustained attack. An inline runtime guardrail that classifies prompts and responses — like Guardion's prompt-defense models — blocks jailbreaks and prompt injection before they reach GPT OSS 120B.
Harden GPT OSS 120B in production
Guardion's runtime guardrails block jailbreaks and prompt injection before they reach any model — with sub-130ms policy decisions.
Request a demo